You are not logged in.
Hi!
chromium gets slower and slower over time until it's basically unusable after a week (while the main process' memory footprint continues to grow). Even just typing in the omnibox is slow. This is actually the very first you'll notice.
# sudo strace -wcp 7813
strace: Process 7813 detached
% time seconds usecs/call calls errors syscall
------ ----------- ----------- --------- --------- ----------------
58.85 0.569705 9 59334 59322 recvmsg
20.59 0.199350 12 16399 sendto
10.11 0.097824 9 9909 poll
2.61 0.025239 10 2401 read
1.85 0.017879 14 1249 madvise
1.47 0.014223 10 1366 openat
1.11 0.010733 7 1370 close
1.04 0.010099 22 456 write
0.87 0.008453 8 1023 stat
0.80 0.007756 7 1026 fstat
0.32 0.003059 10 288 102 futex
0.26 0.002491 7 341 lseek
0.08 0.000766 28 27 munmap
0.01 0.000120 10 11 writev
0.01 0.000083 83 1 fallocate
0.01 0.000078 15 5 sendmsg
0.01 0.000074 10 7 getpid
0.00 0.000046 22 2 mmap
0.00 0.000043 10 4 fcntl
0.00 0.000011 11 1 unlink
0.00 0.000010 10 1 ftruncate
------ ----------- ----------- --------- --------- ----------------
100.00 0.968044 95221 59424 totalBasically all recvmsg calls fail. Is this normal? And it's always the same few sockets
# sudo strace -p 7813 2>&1 | grep recvmsg | sort | uniq
^Crecvmsg(31,
recvmsg(31, {msg_namelen=0}, 0) = -1 EAGAIN (Resource temporarily unavailable)
recvmsg(31, {msg_name=NULL, msg_namelen=0, msg_iov=[{iov_base="U\2,\364\331S\372\20\3\4\4\0\0\0\0\0\0\0\0\4\4\4\4\4\0\0\3\37%\2\0\0", iov_len=4096}], msg_iovlen=1, msg_controllen=0, msg_flags=0}, 0) = 32
recvmsg(32, {msg_namelen=0}, 0) = -1 EAGAIN (Resource temporarily unavailable)# ls -l /proc/7813/fd/3{1,2}
lrwx------ 1 tcn tcn 64 Nov 23 08:19 /proc/7813/fd/31 -> 'socket:[57168]'
lrwx------ 1 tcn tcn 64 Nov 23 08:19 /proc/7813/fd/32 -> 'socket:[57169]'
# grep -e 57168 -e 57169 /proc/net/unix
0000000010819d23: 00000003 00000000 00000000 0001 03 57168
00000000e0dc0099: 00000003 00000000 00000000 0001 03 57169Anybody else with this problem? Any idea how to track this down further?
chromium 78, but have had this issue for a long time.
Last edited by tcn (2019-11-23 14:33:01)
Offline
Hello! I have the same problem.
I have no way to tell if that's related, but I discovered it while investigating about https://bugs.chromium.org/p/chromium/is … ?id=995368
Do you have the same issue as well?
# strace -wcp 87512
strace: Process 87512 attached
^Cstrace: Process 87512 detached
% time seconds usecs/call calls errors syscall
------ ----------- ----------- --------- --------- ----------------
50.86 11.753437 16 725176 725092 recvmsg
29.05 6.713140 55 120864 poll
7.48 1.727458 16 106265 365 futex
4.95 1.143050 15 73480 read
3.75 0.867271 22 37861 write
1.12 0.258473 154 1675 sendto
0.76 0.176758 31 5635 openat
0.66 0.152780 42 3577 madvise
0.49 0.113963 18 6028 close
0.34 0.079571 23 3381 fstat
0.30 0.070032 20 3383 stat
0.13 0.030602 77 393 munmap
0.08 0.019274 17 1127 lseek
0.01 0.002084 173 12 writev
------ ----------- ----------- --------- --------- ----------------
100.00 23.107892 1088857 725457 totalThe fds are 31 and 35 for me
# ls -l /proc/87512/fd/31
lrwx------ 1 arch users 64 Nov 28 10:18 /proc/87512/fd/31 -> 'socket:[481552]'# ls -l /proc/87512/fd/35
lrwx------ 1 arch users 64 Nov 28 10:18 /proc/87512/fd/35 -> 'socket:[481554]'# grep -e 481552 -e 481554 /proc/net/unix
0000000029c70ef8: 00000003 00000000 00000000 0001 03 481554
00000000ca047af5: 00000003 00000000 00000000 0001 03 481552I've done the same investigation steps as you but I'm not sure where you were going.
Offline
Starting with a new profile (I renamed my current profile folder) fixes both issues:
strace: Process 90778 attached
^Cstrace: Process 90778 detached
% time seconds usecs/call calls errors syscall
------ ----------- ----------- --------- --------- ----------------
99.11 15.064985 273908 55 poll
0.79 0.119669 119669 1 restart_syscall
0.09 0.013649 40 337 310 recvmsg
0.00 0.000471 39 12 futex
0.00 0.000374 46 8 read
0.00 0.000214 53 4 sendto
0.00 0.000194 194 1 close
0.00 0.000152 152 1 mmap
0.00 0.000081 20 4 writev
0.00 0.000074 36 2 stat
0.00 0.000037 37 1 fcntl
0.00 0.000035 35 1 munmap
------ ----------- ----------- --------- --------- ----------------
100.00 15.199936 427 310 totalOffline
So pesky. Maybe really should just switch to a mac...
One debugging step further:
# readlink /proc/7625/fd/3{1,2}
socket:[41838]
socket:[41839]
# lsof -p 7625 -a -d 31 -d 32
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
chromium 7625 tcn 31u unix 0x00000000b350e7b4 0t0 41838 type=STREAM
chromium 7625 tcn 32u unix 0x0000000024245bbf 0t0 41839 type=STREAM
# ss -p | grep -e 41838 -e 41839
u_str ESTAB 0 0 @/tmp/.X11-unix/X0 52964 * 41838 users:(("Xorg",pid=6569,fd=68))
u_str ESTAB 0 0 * 41838 * 52964 users:(("chromium",pid=7625,fd=31))
u_str ESTAB 0 0 * 41839 * 52966 users:(("chromium",pid=7625,fd=32))
u_str ESTAB 0 0 @/tmp/.X11-unix/X0 52966 * 41839 users:(("Xorg",pid=6569,fd=72)) So, it's apparently Xorg that doesn't return data. I'm guessing that fits the secondary problem that resizing windows is at times very sluggish. Or even impossible (for a while when starting off the browser and restoring the session).
Last edited by tcn (2020-01-05 18:53:44)
Offline